My project idea is to make a drone that can recognize distress signals at an accident scene and detect their location. I have referenced the following Edge Impulse documentation for your project:
- Location Identification using Sound - Expert Projects
- Detecting Worker Accidents with Audio Classification - Expert Projects
However, I am unsure about how to teach the drone to recognize the direction from which the sound is coming.
Can I preprocess each dataset by adding direction information, such as east, west, south, north, up, and down, resulting in six times more datasets for training? Or is there another method that I can use? I would appreciate any advice you can give me.